Health has a social gradient, i.e. people with lower socio-economic status (SES) have more health related problems than people in the other part of the SES hierarchy 12345. Further SE inequalities seem to increase in Western European countries 67. There may be three possible pathways: increased risk of socially patterned unhealthy lifestyle in lower SES groups 89, unequal access to health care 1011, and finally various exposure to material deprivation and a stressful psychosocial environment 1213141516.

Inequalities in health across ethnic groups, in terms of both morbidity and mortality, have been documented in the United States 1718192021 and the United Kingdom 222324252627, both across social class and ethnicity. In the United States the mortality rate for black population has been at least 60 percent higher than for the white population, and trend seems to continue 2829. A number of studies have shown that most British ethnic minorities have higher mortality and morbidity rates compared to the majority ethnic British population 222330. The reported mortality in immigrants with South-Asian descent in the UK caused by coronary diseases, is 40 percent higher than that of Europeans 31. This seems to be linked to the higher prevalence of diabetes in this group 323334. In general, mortality from both cardiovascular diseases and diabetes is more common in immigrant communities especially from South Asia 3536 and is linked with the poor SE conditions 37383940. Similar results have been found for South-Asians immigrants in other countries 324142.

A SE gradient has been reported for health inequalities in Norway 4344454647. A study on health inequalities by SES among men has suggested an increased gradient in health inequalities for the period of 1980 to 1990. The inequality pattern appeared to be closer to the average in other European countries 48. Similarly Norwegian women has shown SE gradient in self-reported health 49.

Inequalities in health among different ethnic groups in Norway are not extensively studied, despite the fact that earlier surveys and official reports had indicated a different morbidity pattern for immigrants in Norway compared to the native population 50. Further, it has been reported that immigrants from non-Western countries perceive their health as poor and have high morbidity due to psychological distress and diabetes 515253.

Self-rated health (SRH) is an important element of the clinical investigation and public health surveillance 5455. It is a subjective appraisal of health and a powerful predictor of survival, functional decline, future morbidity, and subsequent health service utilization. It's validity as a measure of health outcome has been explored after controlling for a variety of physical, socio-demographic, and psychosocial health indices 56. Moreover, its importance in predicting mortality across SE categories even after adjusting for objective disease, has been reported in earlier studies 5758. For that reason, it has been widely used as a health outcome measure in studies investigating SE inequalities in health. Further it has been reported that immigrants in comparison to the native populations in European context often rate their health as poor 59.

Psychological distress is a measure of mental health, represented by symptoms of anxiety, depression and somatization. Studies from the Western countries have shown a social gradient; higher SES is associated with lower rate of psychological distress. Moreover, it has been suggested that differences in SES may explain differences in psychological distress between certain groups of immigrants and the native population in Western countries 60.

Self-reported diabetes was another outcome variable included in the study. The association of many vascular diseases and their risk factors with SES has been well described 6162. Certain risk factors implicated in the development of diabetes are also known to be associated with SES. Obesity, low physical activity, smoking, and low birth weight have all been described as risk factors for type 2 diabetes. In Western societies these factors are associated with lower SES. Thus an inverse relationship would be expected between the prevalence of type 2 diabetes and SES 63. A few ecological studies have also described, an inverse relation between incidence of type 2 diabetes and relative affluence of the towns 6364. In addition to this knowledge, we also know that immigrants from South-Asia have higher prevalence of diabetes type 2 and coronary heart disease than ethnic Europeans 3241. This high prevalence has also been reported for the immigrants from South-Asia in Norway 5265.

Because of the already reported poor SRH, high rates of psychological distress and diabetes among non-Western immigrants in Norway, and additionally their possible relation to SES, the aim of the present study is to investigate to which extent differences in these measures of health between Pakistani immigrants in Oslo may be explained by differences in SES.

In Norway, immigrants with Pakistani background constitute the largest non-Western immigrant group. To the best of our knowledge no previous study has particularly explained inequality in health for this ethnic group from the vantage of SE gradient. Therefore, the objective of our study is to investigate the health inequalities with respect to the selected indicators from the vantage of SE factors between ethnic Pakistanis and Norwegians in Oslo, Norway.

The participants with ethnic Pakistani background (mean age 42.1 ± 10.8) were significantly younger (p < 0.001) than ethnic Norwegians, (mean age 50.1 ± 16.1). More than 50% of the Pakistani participants were in the middle age (i.e. 40–45 years) while Norwegians were more evenly distributed across all the age groups. All other indices of socio demographic and economic characteristics, like gender, marital status, and education, level of employment, occupational class and household income differed significantly among the participants from these two different ethnic communities (Table 1).

The level of unemployment was four times higher among the Pakistanis compared to the Norwegians. 66% of the ethnic Pakistanis were in the low income category, and 7% were in the high income group compared to 39% in the high income category in the Norwegians. These socio-demographic differences are in accordance with the already existing statistics on Pakistani immigrants in Norway 7172.

Ethnic Pakistanis reported higher prevalence of poor SRH, diabetes and distress compared to the Norwegian population. This difference was significant both unadjusted and adjusted for age and sex (p < 0.001) (Table 2).

Poor health was more often reported by ethnic Pakistanis for each subcategory of SE parameters (Table 3). An inverse social gradient was observed for Norwegians in respect to SRH, diabetes and distress, where as no such trend was detected for the Pakistanis. Levels of education did not show any consistent relationship neither with SRH nor diabetes among the Pakistanis; where as an opposite trend was observed for distress. Each indicator of health such as SRH, diabetes and distress showed consistent inverse relationship with income for the Norwegians. As for Pakistanis this was weak or even not existing in case of Pakistani respondents. All differences between Norwegians and Pakistanis with respect to the effect of SES, was tested for interaction by logistic regression analysis (data not shown). The following interactions were statistically significant: ethnicity* education with respect to SRH (p = 0.001) and distress (p < 0.001), ethnicity*employment with respect to SRH (p = 0.003) and distress (p < 0.001). In other words, there was a difference in the effect of education and employment on SRH and distress for the ethnic Pakistanis and Norwegians.

SE conditions, in addition to age and sex modified the odd ratios for SRH and distress but not for diabetes among the ethnic Pakistanis in the multivariate regression model (Table 4). Following the different models employed in the regression analysis it is apparent that employment status appeared to have the strongest impact on SRH and distress and to a lesser degree on diabetes.

Our results confirm the overall poor health reported by the ethnic Pakistanis compared to the Norwegians, irrespective of SE status. The economical gradient is less marked for Pakistanis, and even an opposite trend was observed for distress. For example, individuals with a reported higher education and or higher household income have shown an association with good health among the Norwegians, where as inconsistent result was noted for the Pakistanis.

Possible explanations for the disparity in the observed associations of "health" with education and income for the ethnic Pakistanis and the Norwegians may entail that the ethnic Pakistanis at large belong to the low levels of education and income group. Given the small number of participants from Pakistan belonging to predominantly high education and income strata, it is possible that we do not observe any association of the social gradient with health although the relationship exists. The other possible explanation is that individuals with higher education from the ethnic Pakistani families were not successful in obtaining an employment that may correspond with their educational background. Therefore higher education for the Pakistani's did not result in improved economy and thereby health. Moreover underemployment may interfere with self esteem which may result in stress and depression with an obvious consequence on health.

Another factor that could contribute to the lack of positive association between education and health in the Pakistanis is inaccuracy in reporting education. In a recently conducted study in Oslo, the validity of self-reported education by immigrant communities has been discussed 73. The conclusion was that in studies based on self-reporting, the tendency to over report education due to social desirability could not be ignored.

Among the SE indices, employment appears to have the maximum impact in explaining higher self reported morbidity among the Pakistanis. This may suggest that being employed, though not necessarily with a high income, has a positive impact on health. This is in accordance with numerous studies showing the negative effect of unemployment 7475767778. The observed unemployment in our study is in accordance with the already existing statistics in Norway. For Pakistanis reported employment rate is 44% compared to 69.3% for the total population. Moreover, described employment pattern for this ethnic group indicated that either they were self-employed with a considerable engagement in hotels and restaurants, or they were employed within elementary occupations 71. In addition, it has been reported that immigrants do not perform as well on the labor market as natives with similar characteristics and a large proportion of immigrants from non-Western countries is characterized as self-employed marginalized, even when controlling for observed and unobserved individual characteristics 79.

The design of the study was cross-sectional and we know that inherent problems of a cross-sectional design is that the outcome (in this case self-reported health) and the exposure (in this case SE conditions) are collected simultaneously and thereby cannot predict the causality. Moreover, cross-sectional studies pay little attention to the information bias emerging from the dependent error, which means a possible correlation between the degree of error in measured exposure and measured outcome. Hence, it is possible that estimated associations between self-reported health and SE factors is falsely inflated or deflated in our study 80.

Self-reporting was the mode of data collection for both health and socio-demographic variables in this study. The data collected by self-reporting has often raised the concerns over its validity. However, self-reported health is widely used in European studies 818283 and in American studies 8485. Self-reported health appeared to be an important independent predictor of all causes of mortality 8687. However, one cannot rule-out the possibility that cultural differences between ethnic groups may imply that they perceive their physical and mental health differently 88. This could be the reason for the differences observed in the self-reported health measures across SE indicators for the ethnic Pakistanis in this study. In the context of our study, it seemed that while reporting their health ethnic Pakistanis were contextualizing their prevailing social condition in the Norwegian society. It might be due to this reason that constructs of health such as SRH and distress related to integrated personal perceptions and overlapping in their abilities to incorporate multiple domains of health such as psychosocial well-being, social functioning, and positive affects showed more obvious relation to the SES among Pakistanis. The same explanation could be used to describe the results of our study with regards to diabetes. A discrepancy in perception between reported diabetes prevalence and longstanding illness has been described for ethnic Pakistanis in the U.K. It was further reported that this group does not consider any illness as a longstanding illness unless it cause severe functional disability 89.

Another methodological challenge was related to the low participation rate for both the ethnic Pakistanis and Norwegians in our study. Low participation in epidemiological studies may threaten the validity and generalizability of the results due to the possibility of selective participation 90. Though in our case, socio-demographic data provided by the central national register showed that the socio-demographic characteristics of participants were nearly the same as those of the non-participants, with slight deviation for ethnic minorities 67.

In our study, we used HSCL-10 as an investigating tool for psychological distress. Even though, it was translated into Urdu, it was not validated for the ethnic Pakistanis. Therefore, the results based on HSCL-10 for distress will have to be interpreted with caution.

Inequalities in health between the ethnic Norwegians and Pakistanis were reduced, but not eliminated, when a number of socio-demographic factors were adjusted in a multivariate model. The reduction was strongest for SRH and psychological distress, where as it was weaker for diabetes. The reason for this difference could be that diabetes to a lesser degree than the other measures of health is influenced by the stressors related to migration. Unemployment seemed to be the most important explanatory factor for the difference in health. Given the context of a social-welfare state like Norway, public health policies should be developed to cope against the discrimination in employment for ethnic minorities with a view to promote health.

The author(s) declare that they have no competing interests.

HR conducted the basic study. OS participated and helped during the all phases of this study. AH contributed with the intellectual discussions and some inputs in draft. ID contributed with statistical help. BC read the script for expert comments. NA also contributed with academic discussions.